Create Machine learning model for public bike parking lots predictions

Create Machine learning model for public bike parking lots predictions#

After Worked on data analytics, we will get started to build up machine learning models

import numpy as np
import pandas as pd
import os
from sklearn.model_selection import train_test_split
from IPython.display import display



pd.set_option('display.max_rows',5)
pd.set_option('display.max_columns', None)  # 显示所有列
pd.set_option('display.width', None)  # 自动调整宽度

li = []
for dirname, _, filenames in os.walk('./data/'):
    for filename in filenames:
        df = pd.read_csv(dirname+filename, index_col=None, header=0)
        li.append(df)

frame = pd.concat(li, axis=0, ignore_index=True)

frame

	_id	station_no	station_name	station_name_en	district	district_en	total	rentAvailable	returnAvaiable	updated_time	info_time
0	66477f00ffd73e577b8e8ee9	500101001	YouBike2.0_捷運科技大樓站	YouBike2.0_MRT Technology Bldg. Sta.	大安區	Daan Dist.	28	2	26	2024-05-17 23:58:23	2024-05-18 00:00:00
1	66477f00ffd73e577b8e8eea	500101002	YouBike2.0_復興南路二段273號前	YouBike2.0_No.273， Sec. 2， Fuxing S. Rd.	大安區	Daan Dist.	21	4	17	2024-05-17 23:58:23	2024-05-18 00:00:00
...	...	...	...	...	...	...	...	...	...	...	...
814260	664f66d4ffd73e577be875f2	500119082	YouBike2.0_臺大四號館東北側	YouBike2.0_NTU Building No.4(Northeast)	臺大公館校區	NTU Dist	10	6	4	2024-05-23 23:53:23	2024-05-23 23:55:00
814261	664f66d4ffd73e577be875f8	500119088	YouBike2.0_臺大黑森林西側	YouBike2.0_NTU Black Forest(West)	臺大公館校區	NTU Dist	20	2	18	2024-05-23 23:53:23	2024-05-23 23:55:00

814262 rows × 11 columns

Before we start building machine learning model#

We have several things to do when build up model

drop unused columns
data type transformations
leaving out the outliers
split data into training set and test set

Drop unused columns#

In the last page, we have identified the data with the type. we start to drop columns with serial numbers or their station name.

frame.drop(columns=['_id', 'station_name','station_name_en','district','district_en'],inplace=True)

frame

	station_no	total	rentAvailable	returnAvaiable	updated_time	info_time
0	500101001	28	2	26	2024-05-17 23:58:23	2024-05-18 00:00:00
1	500101002	21	4	17	2024-05-17 23:58:23	2024-05-18 00:00:00
...	...	...	...	...	...	...
814260	500119082	10	6	4	2024-05-23 23:53:23	2024-05-23 23:55:00
814261	500119088	20	2	18	2024-05-23 23:53:23	2024-05-23 23:55:00

814262 rows × 6 columns

Leaving out error values#

frame = frame.loc[(frame['total'] == frame['rentAvailable'] + frame['returnAvaiable']),:]

frame.drop(columns='station_no',inplace=True)

frame

/tmp/ipykernel_1839/1947658609.py:3: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  frame.drop(columns='station_no',inplace=True)

	total	rentAvailable	returnAvaiable	updated_time	info_time
0	28	2	26	2024-05-17 23:58:23	2024-05-18 00:00:00
1	21	4	17	2024-05-17 23:58:23	2024-05-18 00:00:00
...	...	...	...	...	...
814260	10	6	4	2024-05-23 23:53:23	2024-05-23 23:55:00
814261	20	2	18	2024-05-23 23:53:23	2024-05-23 23:55:00

707282 rows × 5 columns

## convert datetime strings to datetime formats
frame['info_time'] = pd.to_datetime(frame['info_time'],format='%Y-%m-%d %H:%M:%S')
frame['updated_time'] = pd.to_datetime(frame['updated_time'],format='%Y-%m-%d %H:%M:%S')

## calculate time difference between station updated time and current time
timeDiff = (frame['info_time'] - frame['updated_time']) / pd.Timedelta(minutes=1)

frame['TimeDiff'] = timeDiff
frame['hours'] = frame['info_time'].dt.hour

/tmp/ipykernel_1839/1608152123.py:2: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  frame['info_time'] = pd.to_datetime(frame['info_time'],format='%Y-%m-%d %H:%M:%S')
/tmp/ipykernel_1839/1608152123.py:3: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  frame['updated_time'] = pd.to_datetime(frame['updated_time'],format='%Y-%m-%d %H:%M:%S')
/tmp/ipykernel_1839/1608152123.py:8: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  frame['TimeDiff'] = timeDiff
/tmp/ipykernel_1839/1608152123.py:9: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  frame['hours'] = frame['info_time'].dt.hour

## judge if current time is workday or not
def isWorkHour (timeSeries):
    work_day = timeSeries.weekday() < 5
    work_hour = timeSeries.hour >= 8 & timeSeries.hour < 19
    return work_day & work_hour

workday = frame['info_time'].apply(isWorkHour)

frame['isWorkDay'] = workday

frame

/tmp/ipykernel_1839/2541519960.py:9: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  frame['isWorkDay'] = workday

	total	rentAvailable	returnAvaiable	updated_time	info_time	TimeDiff	hours	isWorkDay
0	28	2	26	2024-05-17 23:58:23	2024-05-18 00:00:00	1.616667	0	False
1	21	4	17	2024-05-17 23:58:23	2024-05-18 00:00:00	1.616667	0	False
...	...	...	...	...	...	...	...	...
814260	10	6	4	2024-05-23 23:53:23	2024-05-23 23:55:00	1.616667	23	True
814261	20	2	18	2024-05-23 23:53:23	2024-05-23 23:55:00	1.616667	23	True

707282 rows × 8 columns

## Add ratio between rentAvailable and total
ratio = frame['rentAvailable'] / frame['total']

frame['ratio'] = ratio

/tmp/ipykernel_1839/133128789.py:4: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  frame['ratio'] = ratio

## remove columns with date data type
frame.drop(columns=['info_time','updated_time'], inplace=True)

/tmp/ipykernel_1839/3284268876.py:2: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  frame.drop(columns=['info_time','updated_time'], inplace=True)

import seaborn as sns
import matplotlib.pyplot as plt

corr_matrix = frame.corr()

# 繪製 heatmap
plt.figure(figsize=(10, 8))
sns.heatmap(corr_matrix, annot=True, fmt=".2f", cmap='coolwarm', linewidths=0.5)
plt.title('Correlation Heatmap')
plt.show()

_images/97236d0f9d4b94224362a873553e5432898940048477f7d7998a1f5cf461c698.png

frame1 = frame.copy()
frame1.drop(columns=['returnAvaiable','isWorkDay'],inplace=True)
corr_matrix = frame1.corr()

# 繪製 heatmap
plt.figure(figsize=(10, 8))
sns.heatmap(corr_matrix, annot=True, fmt=".2f", cmap='coolwarm', linewidths=0.5)
plt.title('Correlation Heatmap')
plt.show()

_images/4865381dc984cccd75d945c11c755f4779acdc5c3079a9b80674902cd632e8e5.png

## split training set and test set
# 分割特徵和標籤
X = frame.drop('rentAvailable', axis=1).loc[:,['total', 'TimeDiff','hours','ratio','isWorkDay']]  # 特徵
y = frame['rentAvailable']  # 標籤

# 使用 train_test_split 將資料分割成訓練集和測試集
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=42)

# 檢查分割後的資料集大小
print(f"X_train shape: {X_train.shape}")
print(f"X_test shape: {X_test.shape}")
print(f"y_train shape: {y_train.shape}")
print(f"y_test shape: {X_test.shape}")

X_train shape: (495097, 5)
X_test shape: (212185, 5)
y_train shape: (495097,)
y_test shape: (212185, 5)

## Loss Function
def rmsle(predicted, actual):
    """
    Calculate the Root Mean Squared Logarithmic Error between predicted and actual values.
    
    Parameters:
    predicted (numpy array): Predicted values
    actual (numpy array): Actual values
    
    Returns:
    float: RMSLE value
    """
    # Ensure the input arrays are numpy arrays
    predicted = np.array(predicted)
    actual = np.array(actual)
    
    # Calculate the logarithms of the predicted and actual values plus one
    log_predicted = np.log1p(predicted)
    log_actual = np.log1p(actual)
    
    # Compute the squared differences between log predictions and log actuals
    squared_log_diff = np.square(log_predicted - log_actual)
    
    # Compute the mean of the squared log differences
    mean_squared_log_diff = np.mean(squared_log_diff)
    
    # Compute the RMSLE
    rmsle_value = np.sqrt(mean_squared_log_diff)
    
    return rmsle_value

from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_squared_error, mean_absolute_error, r2_score

# 建立並訓練線性回歸模型
lm = LinearRegression()
lm.fit(X_train, y_train)
y_pred = lm.predict(X_test)
# predicted number must be greater than 0
y_pred = np.clip(y_pred,0,None)
# 計算均方誤差 (MSE)
mse = mean_squared_error(y_test, y_pred)

# 計算均絕對誤差 (MAE)
mae = mean_absolute_error(y_test, y_pred)

# 計算 R 平方 (R² Score)
r_squared = r2_score(y_test, y_pred)

# 計算調整後的 R 平方
n = X_test.shape[0]
p = X_test.shape[1]
adj_r_squared = 1 - (1 - r_squared) * (n - 1) / (n - p - 1)

# 印出模型績效
print("Mean Squared Error (MSE):", mse)
print("Mean Absolute Error (MAE):", mae)
print("R Squared (R²):", r_squared)
print("Adjusted R Squared (Adjusted R²):", adj_r_squared)
print("RMSLE score: ", rmsle(y_pred,y_test))
print("Linear model scores: ", lm.score(X_test,y_test)*100)
# 輸出 y_pred
y_pred

Mean Squared Error (MSE): 13.69821432798849
Mean Absolute Error (MAE): 2.1776454006749235
R Squared (R²): 0.8300403407405614
Adjusted R Squared (Adjusted R²): 0.8300363356396971
RMSLE score:  0.48873312679312847
Linear model scores:  81.64566520126644

array([10.51909957,  6.17997853, 14.51628116, ...,  7.70377745,
       20.95194595,  1.61564677])

Create Machine learning model for public bike parking lots predictions

Contents

Create Machine learning model for public bike parking lots predictions#

Before we start building machine learning model#

Drop unused columns#

Leaving out error values#